A guide to child nonverbal IQ measures. American Journal of

advertisement
Clinical Focus
A Guide to Child Nonverbal IQ Measures
Laura S. DeThorne
Barbara A. Schaefer
Pennsylvania State University, University Park
This guide provides a basic overview of 16
child nonverbal IQ measures and uses a set of
specified criteria to evaluate them in terms of
their psychometric properties. In doing so, the
goal is neither to validate nor to criticize current
uses of IQ but to (a) familiarize clinicians and
investigators with the variety of nonverbal IQ
T
he use of IQ scores has become relatively commonplace within the clinical and research practices of
speech-language pathology. From a clinical standpoint, speech-language pathologists in the public schools
are often encouraged to consider IQ when making decisions regarding treatment eligibility (Casby, 1992;
Whitmire, 2000). From a research perspective, IQ measures
are often administered to define the population of interest, as
in the case of specific language impairment (Plante, 1998;
Stark & Tallal, 1981). One distinction across IQ measures
that appears particularly relevant to the field of speechlanguage pathology is whether the measure of interest is
considered verbal or nonverbal in nature. Nonverbal
intelligence tests were designed to measure general cognition without the confound of language ability. The major
impetus for the development of nonverbal IQ measures
appeared to come from the U.S. military during World War
I. As summarized in McCallum, Bracken, and Wasserman
(2001), the military used group-administered IQ measures to
place incoming recruits and consequently needed a means to
assess individuals that either were illiterate or demonstrated
limited English proficiency. Once nonverbal assessment
tools were developed, their use expanded to include additional populations, such as individuals with hearing loss,
neurological damage, psychiatric conditions, learning
disabilities, and speech-language impairments. Because the
practice of speech-language pathology centers on such
special populations, the present article focuses exclusively
on measures of nonverbal IQ.
The distinction between verbal and nonverbal IQ was
not originally based on empirically driven theory; rather,
Wechsler’s original IQ scales differentiated verbal from
performance (nonverbal) primarily for practical reasons
(McGrew & Flanagan, 1998, p. 25). Subsequent theorists
have applied factor analytic methods to a variety of
measures currently available, (b) highlight some
of the important distinctions among them, and
(c) provide recommendations for the selection
and interpretation of nonverbal IQ measures.
Key Words: assessment, cognition, language
cognitive measures to derive empirically based models of
intelligence. For example, Carroll (1993) conducted factor
analyses of 477 data sets to develop his three-stratum
theory of intelligence. Stratum III includes a general factor,
often referred to as g, which is thought to contribute in
varying degrees to all intellectual activities. The second
stratum contains 8 broad abilities, such as fluid intelligence
or visual perception. These broad abilities are then further
divided into 70 narrow cognitive abilities that contribute to
Stratum I. McGrew and Flanagan (1998) have integrated the
three-stratum theory with the theory of fluid and crystallized
intelligences, associated with Horn and Cattell (1966), to
create the Cattell–Horn–Carroll theory of cognitive abilities
(McCallum, 2003, p. 64). Based on this theory, McGrew
and Flanagan (1998) have suggested that verbal IQ
represents the construct of crystallized intelligence but that
“nonverbal IQ” is not a valid construct in and of itself:
“There is no such thing as ‘nonverbal’ ability—only
abilities that are expressed nonverbally” (p. 25).
Although questions regarding the nature of intelligence
and how it is best measured are beyond the scope of this
article, the Cattell–Horn–Carroll theory of cognitive
abilities (McGrew & Flanagan, 1998) is offered as a
theoretical framework within which to speculate about the
nature of individual IQ measures. Previous published work
has provided thorough discussion of the psychometric
properties and general issues related to standardized
assessment (e.g., McCauley, 2001; McCauley & Swisher,
1984a, 1984b). The present review intends to expand on
such work by applying the information to specific nonverbal IQ measures and by including a discussion of how
individual tests may relate to broad cognitive abilities. The
specific aims of the present article are to (a) familiarize
clinicians and investigators with the variety of nonverbal
IQ measures currently available, (b) highlight some of the
American Journal of Speech-Language Pathology • Vol. 13 • 275–290 • November 2004 • © American Speech-Language-Hearing Association
1058-0360/04/1304-0275
DeThorne & Schaefer: Nonverbal IQ
275
important distinctions among them, and (c) provide
recommendations for the selection and interpretation of
nonverbal IQ measures.
Overview
We undertook a review of all mainstream IQ measures
currently on the market and selected for this review those
that met five specific criteria. First, each measure had to be
marketed or commonly used as a measure of general
cognitive functioning. Second, each included measure had
to provide a standardized score of nonverbal ability. By
nonverbal, we mean that the tests rely heavily on visuospatial skills and do not require examinees to provide
verbal responses. Some IQ measures, such as the Cognitive
Assessment System (CAS; Naglieri & Das, 1997) and the
Woodcock–Johnson Tests of Cognitive Abilities—III
(Woodcock, McGrew, & Mather, 2001), include nonverbal
components but do not organize them into a composite or
scaled score. Such measures were not included given that
interpretation at the subscale level is generally not advised
due to psychometric concerns (McDermott & Glutting,
1997).
Third, in addition to providing a nonverbal measure of
general cognitive functioning, each included IQ measure
was required to be relatively recent, meaning each was
developed or revised within the last 15 years. One exception was made in the case of the Columbia Mental Maturity
Scale—Third Edition (CMMS–3; Burgemeister, Hollander,
& Lorge, 1972). Although this measure was published over
30 years ago, it was included here given its frequent use
within the field of speech-language pathology. The fourth
criterion for test inclusion was suitability of the measure
for preschool or school-age children. Although a number
of the included measures can be administered well into
adulthood, IQ measures that focused exclusively on adults
were omitted. Fifth, each measure had to be developed for
the population at large as opposed to specific groups, such
as individuals with visual impairment or deafness. Table 1
contains a list of 16 IQ measures that met all five specified
criteria. The remainder of the article is devoted to (a)
elaborating on the test information presented in Table 1,
(b) providing an evaluation of each test’s psychometric
properties, and (c) offering recommendations for the
selection and interpretation of nonverbal IQ measures.
Test Information
Age Range
Column 1 lists the ages at which normative data were
collected for the test as a whole or for the nonverbal
section specifically if it differed from the whole.
Verbal Subscore
Column 2 in Table 1 denotes whether each measure
provides a verbal subscore in addition to the nonverbal
components. Although the focus of the present article is on
the nonverbal components of each measure, we thought it
might be useful for readers to know whether the same test
276
provided a verbal subscore as well. Many verbal subscores
are composed of subtests that are analogous to language
measures. For example, the Wechsler Intelligence Scale for
Children—Fourth Edition (WISC–IV; Wechsler, 2003)
includes a Similarities subtest that requires children to
explain how two items are alike, and the Differential
Ability Scales (DAS; C. D. Elliott, 1990a) include a
Naming Vocabulary subtest that requires children to name
objects/pictures. To the extent they reflect language
abilities specifically, verbal subscores may help validate
and/or identify language concerns in particular children.
Verbal subscores can be derived from 5 of the 16 measures
listed in Table 1. Separate from the verbal subscores noted
in column 2, a number of measures presented in Table 1
offer additional batteries, supplemental subtests, or
alternative forms. Such additional components are noted in
column 6, with the subtest descriptions.
Form of Instructions
Column 5 in Table 1 refers to the form of instructions,
either verbal or nonverbal, that is used during test administration. It is not uncommon for measures that are promoted
as nonverbal to rely to varying degrees on verbal instruction. In fact, the nonverbal components of almost all tests
listed in Table 1 include verbal instructions. For example,
even though the child is allowed to respond nonverbally by
pointing, administration of the Picture Completion subtest
from the WISC–IV (Wechsler, 2003) is accompanied by
the instruction, “I am going to show you some pictures. In
each picture there is a part missing. Look at each picture
carefully and tell me what is missing.” Given the use of
verbal instructions, McCallum et al. (2001) have argued
that most “nonverbal tests” are better described as “language-reduced instruments” (p. 8).
Some measures, such as the Kaufman Assessment
Battery for Children (K–ABC; Kaufman & Kaufman,
1983) and the Comprehensive Test of Nonverbal Intelligence (CTONI; Hammill, Pearson, & Wiederholt, 1997),
provide verbal instructions within the administration
procedures, but note that nonverbal instructions could be
used. Such measures are identified by V/NV in column 5
of Table 1. In contrast to measures that simply offer
nonverbal instructions as an administration option, three
measures in Table 1 were specifically normed with
nonverbal instructions: the Leiter International Performance Scale—Revised (Leiter–R; Roid & Miller, 1997),
the Universal Nonverbal Intelligence Test (UNIT; Bracken
& McCallum, 1998), and the Test of Nonverbal Intelligence—Third Edition (TONI–3; Brown, Sherbenou, &
Johnsen, 1997). In such cases, instruction and feedback are
provided nonverbally via gestures, facial expressions,
modeling, and so on. Of course, the use of nonverbal
instructions does not guarantee that language ability will
not influence test performance. Verbal problem-solving
strategies can be used to solve nonverbal problems. In
other words, children can talk themselves through problems that do not require a verbal response. For example, on
recent administration (by the second author) of the UNIT, a
7-year-old girl labeled the figures girl, man, baby, boy on
American Journal of Speech-Language Pathology • Vol. 13 • 275–290 • November 2004
DeThorne & Schaefer: Nonverbal IQ
277
Leiter International
Performance Scale—
Revised (Leiter–R;
Roid & Miller, 1997)
2;0–20;11
4;0–12;6
Kaufman Assessment
Battery for Childrend
(K–ABC; Kaufman &
Kaufman, 1983)
40–60
15–20
Timea
No
No
40
35–50
Yes:
15–30
Verbal
Cluster
at age
3;6–17;11
No
6;0–90;11
2;6–17;11
No
Verbal
subscore
3;6–9;11
Age range
(years;months)
Differential Ability Scales
(DAS; C. D. Elliott, 1990a)
Columbia Mental Maturity
Scale—Third Edition
(CMMS–3; Burgemeister,
Hollander, & Lorge, 1972)
Comprehensive Test of
Nonverbal Intelligence
(CTONI; Hammill, Pearson,
& Wiederholt, 1997)
IQ test
TABLE 1 (Page 1 of 3). Summary of nonverbal IQ measures.
Yes:
foam
shapes,
Yes:
foam
triangles,
photo
cards, &
color form
squares
Yes:
wooden
blocks,
foam
squares,
plastic
cubes,
picture
cards,
and
pencil
No
No
NV
V/NV
V/NV
V/NV
V
Manipulativesb Instructions
Pictorial Analogies—identify relationships in a 2 x 2 picture matrix and
select a response to complete a similar set of relationship pictures
Geometric Analogies—same as Pictorial Analogies using geometric
designs
Pictorial Categories—select 1 out of 5 picture figure(s) that shares a
relationship with 2 stimulus pictures
Geometric Categories—same as Pictorial Categories using geometric
designs
Pictorial Sequences—identify the rule in a progression of picture stimuli
and select a picture to complete the sequence
Geometric Sequences—same as Pictorial Sequences using geometric
designs
Block Building—replicate constructions made by the examiner using
wooden blocks (age 2;6–3;5)
Picture Similarities—place individual picture cards next to similar or
related items within illustrated arrays (age 2;6–5;11)
Pattern Construction—(timed, with untimed alternative) reproduce
pictured designs using foam squares or plastic cubes (age 3;6–17;11)
Copying—copy line drawings made by the examiner or shown in a
picture (age 3;6–5;11)
Matrices—select pictured items to complete rule-based patterns
(ages 6–17;11)
Sequential/Quantitative Reasoning—view pairs of figures/numbers and
determine the missing items (age 6–17:11)
Recall of Designs—draw individual abstract designs after viewing each
briefly (age 6–17;11)
Face Recognition—select matching faces after seeing each for 5 s
(age 4)
Hand Movement—imitate hand movements made by the examiner
Triangles—(timed) replicate pictured geometric configurations using
foam triangles
Matrix Analogies—select pictures or color form squares that complete
the illustrated analogies (age 5–12)
Spatial Memory—identify the positions of pictured items after seeing them
on a target page for 5 s (age 5–12)
Photo Series—put photo cards of a story in chronological order
(age 6–12)
Note. Additional subtests combine to form the Sequential and
Simultaneous Processing Scales, which together form the Mental
Processing Composite; Achievement subtests are also available.
A Spanish version of the test is also available.
Figure Ground—find a variety of pictured objects within increasingly
complex illustrations
Design Analogies—(bonus points on later items) use picture cards to
complete visual analogies (age 6–20)
Complete a series by selecting the picture or figure that does not
belong with the others
Description of nonverbal subtests
$939
$470
$799
$378
$725
Costc
278
American Journal of Speech-Language Pathology • Vol. 13 • 275–290 • November 2004
No
Yes
No
3–94
2–89;11
30
10–15
45
V
V
V
Yes: plastic
V
form board
shapes,
counting
rods, 1-in.
plastic blocks,
pencil
No
Yes: pencils V/NV
No
6–adult
15–30
Raven’s Standard
Progressive Matrices—
Parallel form (SPM–P;
Raven, Raven, &
Court, 1998)
Reynolds Intellectual
Assessment Scales
(RIAS; Reynolds &
Kamphaus, 2003)
Stanford–Binet Intelligence
Scale—Fifth Edition
(SB5; Roid, 2003)
No
3;0–8;11
Pictorial Test of
Intelligence—Second
Edition (PTI–2; French,
2001)
25–30
No
No
Manipulativesb Instructions
5;0–17;11
Timea
Naglieri Nonverbal
Ability Test (NNAT;
Naglieri, 2003)
Verbal
subscore
picture
cards
Age range
(years;months)
Leiter International
Performance Scale—
Revised (continued)
IQ test
TABLE 1 (Page 2 of 3). Summary of nonverbal IQ measures.
Verbal Abstractions—Find the pictured item that is being labeled or
described
Form Discrimination—look at the pictured target item and identify the
associated item from a pictured display
Quantitative Concepts—find the pictured item that is associated with the
verbal description
Note. Test was designed to accommodate eye gaze responses.
Complete 5 sets of 12 illustrated puzzles by selecting from an array of
potential answers (e.g., select the piece that completes the illustrated
pattern)
Note. Additional forms of the Standard Progressive Matrices comprise
the Coloured Matrices and the Advanced Progressive Matrices.
Odd Item Out—(timed) Select picture from array of 5–7 pictures that
makes it different than the others
What’s Missing—(timed) Identify important part of pictured object
that is missing
Object Series/Matrices—solve novel figural problems, sequences of
pictured objects, or matrix-type patterns by pointing to pictured objects
or picking up target objects, such as form board shape, counting rod, or
block (serves as the routing subtest)
Procedural Knowledge—identify common signals and actions
Picture Absurdities—select missing or absurd details in pictures
Quantitative Reasoning—solve premath, arithmetic, algebraic or
functional concepts depicted by selecting blocks, counting rods, or
choosing from illustrations
Form Board—visualize and solve spatial and figural puzzlelike problems
using form board shapes
Form Patterns—(timed) complete patterns by moving plastic form board
shapes into place
Form Completion—assemble foam shapes to match an illustration or pair
pictures of “dissected” objects with pictures of them in whole form
Matching—match foam shapes or picture cards to an array of objects
pictured on an easel (age 2–10)
Sequential Order—select the picture card that best completes the
illustrated array
Repeated Patterns—determine which foam shape or picture card best
completes an illustrated pattern of shapes or objects
Classification—associate foam shapes or picture cards with illustrated
items according to similarities in color, size, function, or shape (age 2–5)
Paper Folding—(bonus points on later items) look at an object pictured on
a card and select the illustration of what that object would look like if
folded (age 6–20)
Note. Also offers an Attention and Memory Battery with 10 different
nonverbal subtests. A Spanish version of the test is also available.
Solve figural matrix items by identifying a pattern and selecting an answer
to complete the pattern.
Note. Parallel forms are available.
Description of nonverbal subtests
$858
$319
£129
$143
$233
Costc
DeThorne & Schaefer: Nonverbal IQ
279
Yes
20–30
25
25
15
30
15–20
Timea
Yes:
diamond
chips
Yes: 1-in.
blocks,
cardboard
puzzle
pieces
Yes: 1-in.
blocks
Yes: 1-in.
blocks
Yes: plastic
chips, 1-in.
cubes,
circular
response
chips, pencil
No
V/NV
V
V
V
NV
NV
Manipulativesb Instructions
Symbolic Memory—model a pictured series of symbols using plastic chips
Spatial Memory—use circular response chips to replicate a pattern of
dots after a brief exposure to a pictured target page
Cube Design—(timed) use cubes to construct a three-dimensional design
that matches a pictured target
Analogic Reasoning—complete a matrix analogy by selecting from four
possible pictured responses
Note. Two additional subtests are available for the test’s extended battery.
Block Design—(timed + bonus points) construct a design with blocks to
match a given target stimulus
Matrix Reasoning—(timed) choose the stimulus to complete the target
matrix or pattern sequence
Block Design—(timed + bonus pts.) construct a design with blocks to
match a given target stimulus
Picture Concepts—identify two to three pictures for different picture arrays
that best go together
Matrix Reasoning—select stimulus to complete target matrix
Note. Supplemental nonverbal subtests are available. A Spanish version
of the test is due for release in the fall of 2004.
Object Assembly—(timed) construct a figure given cardboard puzzle
pieces (age 2;6–3;11)
Block Design—(timed) construct a design with blocks to match a model
presented in the stimulus book
Matrix Reasoning—select from an array of pictured items to complete the
pictured rule-based matrix (age 4–7;3)
Picture Concepts—select the two pictures that go best together from 2
rows of pictured items (age 4–7;3)
Matrices—(timed) select picture that completes rule-based design series
Diamonds—(timed) construct specific designs using single or multiple
diamond-shaped chips
Delayed Response—find an object that has been placed by the examiner
under 1 of 3 cups
Block Span—tap sequences of blocks in the same order as modeled by
the examiner
Note. Depending on examinee’s skill level (as determined via the
Object Series/Matrices routing subtest), not all subtests are administered.
Abstract Figural Problem-Solving—select from an array of pictured items
to complete the pictured rule-based matrix or figure series
Note. Two parallel forms are available.
Description of nonverbal subtests
$250
$799
$799
$235
$579
$299
Costc
a
Note. V = verbal; NV = nonverbal.
Time estimate for nonverbal components only. bRefers to the objects that the examinee is required to manipulate during test administration. cCost as of January 6, 2004, taken from Web
sites or test publisher catalogs in January 2004. dCurrently undergoing revision.
4–85
Wide Range Intelligence
Test (WRIT; Glutting,
Adams, & Sheslow,
2000)
Yes
Yes
6–16;11
2;6–7;3
Yes
No
5;0–17;11
6–89
No
Verbal
subscore
6;0–89;11
Age range
(years;months)
Wechsler Preschool
and Primary
Scale of Intelligence—
Third Edition
(WPPSI–III;
Wechsler, 2002)
Wechsler Abbreviated
Scale of Intelligence
(WASI; Psychological
Corporation, 1999)
Wechsler Intelligence
Scale for Children—
Fourth Edition
(WISC–IV;
Wechsler, 2003)
Test of Nonverbal
Intelligence—Third
Edition (TONI–3;
Brown, Sherbenou,
& Johnsen, 1997)
Universal Nonverbal
Intelligence Test
(UNIT; Bracken &
McCallum, 1998)
Stanford–Binet Intelligence
Scale—Fifth Edition
(continued)
IQ test
TABLE 1 (Page 3 of 3). Summary of nonverbal IQ measures.
the Symbolic Memory subtest to help herself remember the
order of presentation. Another example comes from the
Leiter–R Figure Ground subtest (administered by the first
author) as a 6-year-old boy was scanning an illustration of
clowns in search of a small plus-like design when he
remarked, “I can’t see no band-aid.” Labeling the plus-like
design as a band-aid likely served as a heuristic of sorts
(albeit an unsuccessful one in this case), which could serve
to focus his visual scan on areas within the illustration
where a bandage might occur (i.e., on one of the clown’s
bodies). As these examples illustrate, linguistic ability is
likely to influence all nonverbal IQ measures to some
extent. However, the influence is minimized when both the
instructions and the responses are provided nonverbally.
Description of Nonverbal Components
Column 6 within Table 1 contains a brief description of
the nonverbal tasks or subtests included within each IQ
measure. For most IQ measures, column 6 does not
provide an exhaustive list of all the available subtests. As
previously mentioned, many measures in Table 1 offer a
verbal scale as well. Additional components, such as
associated batteries and subtests that do not qualify as a
core component of either the verbal or nonverbal subscore,
have been noted in column 6.
Embedded within the subtest descriptions in column 6
is information regarding (a) applicable ages and (b) time
constraints. If the age range for a specific subtest does not
differ from the age range provided in column 1, then no
additional age information is provided for that subtest in
column 6. For example, the Leiter–R Visualization and
Reasoning Battery (Roid & Miller, 1997) includes a total
of 10 nonverbal subtests that combine to form the IQ
measure. However, only 7 are administered to any
particular child depending on that child’s age. Note that
the age ranges provided in column 6 for each subtest refer
to the use of that subtest as a core component of the IQ
score, not for its use as a supplemental subtest. The age at
which individual subtests can be administered as supplemental subtests may vary from the information presented
in Table 1.
In addition to information regarding age, it is noted
within the subtest description if the examinee’s response is
subject to time constraints. For example, the Triangles
subtest of the K–ABC (Kaufman & Kaufman, 1983) allots
2 min per item. If the examinee replicates the desired
model but exceeds this time allotment, the item is scored as
incorrect. In addition, some subtests assign bonus points,
depending on the speed with which the task is completed.
Bonus points can exist regardless of whether the subtest is
subject to time constraints. For example, the Paper Folding
subtest of the Leiter–R (Roid & Miller, 1997) does not
require examinees to select the correct response within a
set time frame to receive credit. However, on later items
the examinees can receive additional points for timely
performance. When the examinee’s response has to be
completed within a set time period, the word timed appears
in parentheses at the beginning of the subtest description in
Table 1. Similarly, any subtest that awards bonus points for
280
quick responses is marked by bonus points at the beginning
of the subtest description.
Psychometric Properties
Any overview of nonverbal IQ measures would be
incomplete without consideration of the tests’ normative
construction and psychometric properties. The American
Educational Research Association, American Psychological Association, and National Council on Measurement in
Education worked together to revise their standards for
educational and psychological testing in 1999. The
resulting standards summarize both the general principles
and specific standards underlying sound test development
and use. Test developers hold the responsibility for sound
instrument development, including (a) collection of an
appropriate normative sample and (b) documentation of
adequate reliability and validity. Based in large part on
these professional standards, we have evaluated each
nonverbal measure in terms of its normative sample,
reliability, and forms of validity evidence. A summary of
our evaluation is provided in Table 2 with related information offered in the accompanying text. Our review of
psychometrics is admittedly cursory, and readers are
referred to additional publications for more in-depth
information (e.g., Athanasiou, 2000; McCauley, 2001).
Normative Sample
Developers of norm-referenced tests must identify and
select appropriate normative samples. Normative samples
consist of the participants to whom individual examinees’
performances will be compared. The normative sample of
each measure listed in Table 2 was evaluated in terms of its
(a) size, (b) representativeness, and (c) recency. We rated
each aspect of the normative sample as positive (+) or
negative (–) when information was available. When
adequate information was not provided in the test manual
to evaluate a particular aspect of the normative sample, 0
appears in the relevant column of Table 2.
Size. Reasonably large numbers of children are preferred to minimize error and achieve a representative
sample. To take an extreme example, imagine trying to
find ten 6-year-olds that are representative of all 6-yearolds currently living within the United States! To evaluate
each measure in terms of sample size, we referred to
Sattler’s (2001) recommendation of a 100-participant
minimum for each child age group included in the normative sample. An age group was defined by a 1-year
interval. Of the 16 tests listed in Table 2, 10 successfully
met the criteria for subsample size. Of the 5 measures that
failed, the Leiter–R (Roid & Miller, 1997) included less
than 100 children in groups age 8, 10, and 11–20, with
numbers reaching as low as 45 for some of the 12–17-yearold age groups. Within the normative sample for the
Naglieri Nonverbal Ability Test (NNAT; Naglieri, 2003),
the 16- and 17-year-olds were combined for a subsample
size of 100. Similarly, the 16- and 17-year-olds were
combined in the normative data from the UNIT (Bracken
& McCallum, 1998) for a subsample total of 175, and the
American Journal of Speech-Language Pathology • Vol. 13 • 275–290 • November 2004
DeThorne & Schaefer: Nonverbal IQ
281
0
+
0
–
+
–
–
–
+
–
–
+
Leiter International
Performance Scale—
Revised (Leiter–R;
Roid & Miller, 1997)
Naglieri Nonverbal Ability
Test (NNAT; Naglieri,
2003)
Pictorial Test of
Intelligence—Second
Edition (PTI–2;
French, 2001)
Raven’s Standard
Progressive Matrices—
Parallel form (SPM–P;
Raven, Raven, &
Court, 1998)
Reynolds Intellectual
Assessment Scales
(RIAS; Reynolds &
Kamphaus, 2003)
Stanford–Binet
Intelligence Scale—
Fifth Edition (SB5;
Roid, 2003)
+
+
+
+
+
+
–
Rep
+
Size
Kaufman Assessment
Battery for Children
(K–ABC; Kaufman &
Kaufman, 1983a)
Columbia Mental Maturity
Scale—Third Edition
(CMMS–3;
Burgemeister,
Hollander, & Lorge,
1972)
Comprehensive Test of
Nonverbal Intelligence
(CTONI; Hammill,
Pearson, &
Wiederholt, 1997)
Differential Ability
Scales (DAS;
C. D. Elliott, 1990a)
IQ test
+
+
–
+
+
+
–
+
+
–
Rec
Normative sample
+
0
–
+
–
0
–
–
+
–
Int
–
–
–
0
–
0
–
–
–
–
T-R
np
+
np
+
np
np
np
np
+
np
Inter
Reliability
+
–
–
+
+
–
+
–
+
+
SEM
√
√
np
√
np
√
√
np
√
np
Dev
√
√
√
√
√
√
√
√
√
√
√
√
√
√
np
√
√
√
√
np
FA
Validity
Test comp
TABLE 2 (Page 1 of 2). Evaluation of each measure based on specific psychometric criteria.
√
√
np
√
√
√
np
√
√
np
Group comp
np
np
√
np
np
np
√
np
np
np
Pred ev
Bortner (1965), Bracken & Naglieri (2003),
McCallum (2003)
Athanasiou (2003), Flanagan &
Caltabiano (2003)
McCallum (2003)
Athanasiou (2000), Aylward (1998),
Bracken & Naglieri (2003),
Bradley-Johnson (1997), Drossman &
Maller (2000), McCallum (2003), Van
Lingen (1998)
Aylward (1992), C. D. Elliott (1990b), S. N.
Elliott (1990), Flanagan & Alfonso (1995),
Platt, Kamphaus, Keltgen, & Gilliland
(1991), Reinehr (1992)
Anastasi (1985), Bracken (1985),
Castellanos, Kline, & Snyder (1996),
Coffman (1985), Conoley (1990), Hessler
(1985), Hopkins & Hodge (1984),
Kamphaus (1990), Kamphaus & Reynolds
(1984), Keith (1985), Mehrens (1984),
Page (1985), Sternberg (1983, 1984)
Athanasiou (2000), Bracken & Naglieri
(2003), Farrell & Phelps (2000), Marco
(2001), McCallum (2003), Stinnett (2001)
Egeland (1978), Kaufman (1978)
Additional reviews
282
American Journal of Speech-Language Pathology • Vol. 13 • 275–290 • November 2004
+
+
+
+
+
0
–
+
+
+
–
Rep
+
Size
+
+
+
+
+
+
Rec
–
–
+
+
–
–
Int
0
–
–
–
–
–
T-R
np
+
+
0
np
+
Inter
Reliability
–
+
+
+
+
+
SEM
√
np
np
np
np
√
Dev
√
√
√
√
√
√
Test comp
√
√
√
√
√
√
FA
Validity
np
√
√
√
√
√
Group comp
np
np
np
np
np
np
Pred ev
Stinnett (2003), Widaman (2003)
Hamilton & Burns (2003)
Athanasiou (2000), Bandalos (2001),
Bracken & Naglieri (2003), Farrell &
Phelps (2000), Fives & Flanagan (2002),
McCallum (2003), Young & Assing (2000)
Keith (2001), Linkskog & Smith (2001)
Athanasiou (2000), Atlas (2001), DeMauro
(2001), McCallum (2003)
Additional reviews
Note. Rep = representativeness; Rec = recency; Int = internal; T-R = test–retest; Inter = interrater; Dev = developmental; Test comp = test comparison; FA = factor analysis; Group comp
= group comparison; Pred ev = predictive evidence; + = specified criteria met; – = specified criteria not met; np = such evidence not provided within the text manual; √ = present; 0 =
inadequate information provided by the test manual.
Test of Nonverbal
Intelligence—Third
Edition (TONI–3;
Brown, Sherbenou,
& Johnsen, 1997)
Universal Nonverbal
Intelligence Test
(UNIT; Bracken &
McCallum, 1998)
Wechsler Abbreviated
Scale of Intelligence
(WASI; Psychological
Corporation, 1999)
Wechsler Intelligence
Scale for Children—
Fourth Edition
(WISC–IV;
Wechsler, 2003)
Wechsler Preschool and
Primary Scale of
intelligence—Third
Edition (WPPSI–III;
Wechsler, 2002)
Wide Range Intelligence
Test (WRIT; Glutting,
Adams, & Sheslow,
2000)
IQ test
Normative sample
TABLE 2 (Page 2 of 2). Evaluation of each measure based on specific psychometric criteria.
17- and 18-year-olds were combined within the Wide
Range Intelligence Test (WRIT; Glutting, Adams, &
Sheslow, 2000) for a subsample size of 125. Within the
normative data from the Reynolds Intellectual Assessment
Scales (RIAS; Reynolds & Kamphaus, 2003), children age
11 to 19 years have been combined to form subsamples
ranging in size from 156 to 167 per 2–3-year grouping.
The Raven’s Standard Progressive Matrices—Parallel
Form (SPM–P; Raven, Raven, & Court, 1998) also failed
to meet the sample-size criteria given that its normative
sample of American children was derived by combining
smaller local norms for which individual sample sizes are
not reported.
Representativeness. Related but distinct from the issue
of sample size is the property of representativeness. The
representativeness of a normative sample is critical in
evaluating the usefulness of an instrument for any specific
population. Since these tests were not designed for any
particular group, normative samples should be collected
and stratified to approximate the population percentages
from the U.S. Bureau of the Census (2001). To receive a
positive rating for representativeness in Table 2, a measure
had to meet two criteria. First, the measure’s normative
sample needed to be expressly stratified on the following
variables: race/ethnicity, geographic region, parent
educational level/socioeconomic status, and sex. Second,
the test’s normative sample needed to include children with
disabilities as appropriate given the test’s administration
procedures. For example, it made sense to exclude children
whose visual or motor impairments would have limited
their performance. However, a sample that excludes
children with disabilities as a general rule was not considered representative within our review (see McFadden,
1996, for rationale).
All reviewed measures except for the SPM–P (Raven et
al., 1998) attempted to derive a nationally representative
sample in regard to race/ethnicity, geographic region, parent
educational level/socioeconomic status, and sex. The
measures in Table 2 were more variable in regard to whether
children with disabilities were included. Two measures
received a negative rating for representativeness because
they excluded children with disabilities altogether: the
CMMS–3 (Burgemeister et al., 1972) and the Stanford–
Binet Intelligence Scale (5th ed.; SB5; Roid, 2003). Similarly, the Leiter–R (Roid & Miller, 1997) and the WRIT
(Glutting et al., 2000) received an uncertain rating, identified
as 0 in Table 2, because they did not specify within the
manual whether children with disabilities were included in
the normative sample. Although the second edition of the
Pictorial Test of Intelligence (PTI–2; French, 2001) included
children with disabilities, it received an uncertain rating as
well due to the relatively sparse information it provided on
its normative sample and concerns regarding the extent to
which the sample characteristics differed from U.S. Bureau
of the Census data (see reviews by Athanasiou, 2003;
Flanagan & Caltabiano, 2003).
Recency. In addition to being large and representative,
normative samples should also be relatively recent. The
Flynn (1999) effect has revealed trends in IQ scores across
time such that IQs have increased by 3 points per decade.
Consequently, reliance on instruments with outdated norms
may overestimate youths’ IQ (see Neisser et al., 1996, for
additional discussion). In keeping with Bracken (2000), we
considered the normative data to be acceptably recent if it
had been collected within the last 15 years. Given this
criterion, normative samples from the CMMS–3
(Burgemeister et al., 1972) and the K–ABC (Kaufman &
Kaufman, 1983) were considered out-of-date. Note,
however, that a revised version of the K–ABC is currently
under way.
Reliability
In addition to identifying and selecting appropriate
normative samples, test developers must demonstrate that
their tests have adequate reliability and validity to justify
their proposed use. In classical test theory, reliability, as
measured via correlation coefficients, represents the degree
to which test scores are consistent across items (i.e.,
internal consistency), across time (i.e., test–retest), and
across examiners (i.e., interrater). Accordingly, we
evaluated each measure in terms of internal reliability,
test–retest reliability, and interrater reliability. A fourth
factor, the availability of standard error of measurement for
each summary score, was also considered. Given the focus
of the present article on nonverbal summary scores rather
than subtests, reliability evidence on subtest scores was not
evaluated. Identical to our rating system for the normative
sample, each form of reliability evidence was evaluated as
positive or negative in Table 2. When the test manual did
not provide adequate information for us to determine
whether a specific criterion was met, 0 appears in the
relevant column of Table 2. The reader is referred to
McCauley (2001) for additional information regarding the
constructs of reliability and validity.
Internal reliability. Internal reliability was rated
positively within Table 2 if uncorrected internal reliability
coefficients reached .90 or above for the nonverbal
summary score within each child age group. This criterion
for reliability coefficients is consistent with general
recommendations for test construction (see Bracken, 1988;
McCauley & Swisher, 1984a; Salvia & Ysseldyke, 1998).
Only 5 of the 16 measures met the established criterion for
internal reliability. Of those that did not, the DAS (C. D.
Elliott, 1990a); Leiter–R (Roid & Miller, 1997); RIAS
(Reynolds & Kamphaus, 2003); and WRIT (Glutting et al.,
2000) received uncertain marks because they failed to
provide separate reliability coefficients for each age group.
Although these measures reported reliability coefficients on
combined age groups, such coefficients can mask substantial
variability. Of the measures that provided coefficients at
each age group, 7 failed to consistently meet .90, although
none fell too far from the mark.
Test–retest reliability. The second criterion for test
reliability, in keeping with Bracken (1987), was a test–
retest coefficient of .90 or greater for the nonverbal
summary score at each child age group. In addition, the
retesting had to occur at least 2 weeks, on average, from
the initial test administration. Not a single nonverbal IQ
measure from Table 2 met this criterion. Three measures,
DeThorne & Schaefer: Nonverbal IQ
283
the WRIT (Glutting et al., 2000), Leiter–R (Roid & Miller,
1997), and PTI-2 (French, 2001), received uncertain marks
for test–retest reliability because they reported coefficients
for collapsed age groups, thereby making reliability
estimates for individual age groups impossible to determine. Most measures from Table 2 reported coefficients
from collapsed or select age groups but received a failed
mark (as opposed to uncertain) because the reported
coefficients did not reach .90. Only three measures
presented individual test–retest coefficients for each age
group: the CTONI (Hammill et al., 1997), which reported
coefficients of .79 to .94; the UNIT (Bracken &
McCallum, 1998), which reported coefficients of .83 to
.90; and the WISC–IV (Wechsler, 2002), which reported
coefficients of .85 to .93 across age groups.
Interrater reliability. The third criterion for evaluation
of test reliability was an interrater reliability of .90 or
greater for the nonverbal summary score. We want to be
clear that this form of reliability refers only to the consistency of scoring across examiners and does not evaluate
the consistency of administration across examiners. The
latter is a source of error that should be assessed but was
not reported by any of the reviewed test manuals. Interrater
reliability of scoring procedures for the nonverbal composite was reported for 7 of the 16 measures: the CTONI
(Hammill et al., 1997); PTI–2 (French, 2001); RIAS
(Reynolds & Kamphaus, 2003); TONI–3 (Brown et al.,
1997); Wechsler Abbreviated Scale of Intelligence (WASI;
Psychological Corporation, 1999); WISC–IV (Wechsler,
2003); and Wechsler Preschool and Primary Intelligence
Scale—Third Edition (WPPSI–III; Wechsler, 2002). All of
these received a positive rating for meeting the proposed
criterion except the WASI, which received an uncertain
mark due to its relatively vague statement that interscorer
agreement for the nonverbal subtests “tends to be in the
high .90s” (Psychological Corporation, 1999, p. 129). The
DAS (C. D. Elliott, 1990a) presented interrater reliability
information, but only for particular subtests.
Standard error of measurement. The fourth criterion for
evaluating test reliability refers to the test manual’s
presentation of standard error of measurement (SEM).
Whereas the reliability coefficients can inform us about
group-level consistency, such correlations are not directly
applicable to individual scores. Instead, the SEM is needed.
SEM is a derivative of the reliability coefficient that takes
into account the scale of the measurement used, the
standard deviation, and the internal consistency of the
score. As such, the SEM provides some sense of how much
error is likely encompassed within the obtained score (see
McCauley & Swisher, 1984b, for additional detail). For
example, on the UNIT (Bracken & McCallum, 1998), the
Standard Battery Full-Scale score SEM is approximately 4
points. Consequently if a child received a score of 87, one
could feel reasonably confident that the child’s true score
would be captured within the range of 87 ±4 points (i.e.,
83–91 is the 68% confidence interval). Given that the
derivation of confidence intervals relies on SEM, the final
criterion for test reliability was that the SEM be presented
for the nonverbal summary score at each child age group.
As glimpsed from Table 2, 11 of the 16 tests presented
284
SEM by age for the nonverbal summary score. Of the 5
measures that failed this criterion, the Leiter–R (Roid &
Miller, 1997); RIAS (Reynolds & Kamphaus, 2003); and
WRIT (Glutting et al., 2000) provided SEMs for combined
age groups only. The DAS (C. D. Elliott, 1990a) presented
SEMs for individual age groups, but only for children age
3;6–7;11. The SPM–P (Raven et al., 1998) did not provide
SEM values at all.
Validity
Although evidence for test validity can take many forms,
validity as an overall concept represents the utility of a test
score in measuring the construct it was designed to measure
(McCauley, 2001; Messick, 1989). For IQ tests, developers
should clearly articulate the theoretical grounding of their
instrument and document evidence for validity from a
variety of sources. Table 2 illustrates the validity evidence
presented within each test manual. The presence of each
form of validity evidence within the test manual is indicated
by a √ in Table 2. The letters np appear in the relevant
columns of Table 2 to indicate which forms of validity
evidence were not provided for any particular measure.
Table 2 reviews each measure according to whether it
provided the following most common forms of construct
validity evidence: (a) developmental trends, (b) correlation
with similar tests, (c) factor analyses, (d) group comparisons, and (e) predictive ability. Although demonstration of
internal consistency is another commonly used form of
validity evidence, this evidence was already covered under
the discussion and evaluation of each test’s internal
reliability.
Developmental trends. Demonstrating that a test’s raw
scores increase with age provides necessary, but by no
means sufficient, evidence of a test’s validity. Although
cognitive abilities increase with age, so do many other
qualities—attention, language use, and height, to name just
a few. Perhaps in part because it represents a relatively
weak form of validity evidence, only 8 of the 16 measures
explicitly mention evidence of developmental change in
their discussions of validity (see Table 2).
Correlation with similar tests. In contrast to the
relatively sparse evidence of developmental trends, the
most common form of validity evidence appeared to be
information regarding correlations with other measures.
All 16 tests from Table 2 presented positive correlations
with other measures of general cognition. In addition, there
was a general tendency to report positive correlations with
achievement tests as a form of validity evidence. Of
course, the strength of such intertest correlations in
demonstrating test validity is dependent on the validity of
the test being used for comparison. For example, two tests
developed to assess general cognition might be highly
correlated if both reflect the same confounding variable
(e.g., language skill or test-taking abilities). Given that the
same tests tend to benchmark with each other, intertest
correlation evidence is limited by an air of circularity.
Factor analyses. Another commonly used form of
validity evidence is the presentation of factor analyses,
either confirmatory, exploratory, or both. Without getting
American Journal of Speech-Language Pathology • Vol. 13 • 275–290 • November 2004
into much detail, factor analysis is a statistical procedure
based on item or subscale intercorrelations that is used to
determine whether the test components load on the abilities
of interest (see McCauley, 2001, for additional information).
Such evidence was presented by 14 of the 16 measures. Of
course, the limitation of such evidence is that just because a
test taps a relatively unitary construct or constructs does not
necessarily mean it’s the right construct.
Group comparisons. Given the limitations of the forms
of evidence previously mentioned, additional validity
support is contributed through comparison studies of
specific subgroups—specifically, subgroups with clear
associations with IQ (e.g., children identified as gifted or
with diagnosed cognitive deficits). In other words, the
measure of interest is administered to subgroups of
children with previously established diagnoses to determine how the descriptive data from such subgroups
compare with the test’s normative data. Children with
previously identified cognitive deficits are expected to
score below the normative mean, whereas children
previously identified as gifted would be expected to score
above the normative mean. Of course, the extent to which
subsample means are expected to deviate from the normative mean is less clear, and even clear mean differences can
mask a substantial amount of individual variability.
Nonetheless, 12 of 16 measures provided validity evidence
based on such subgroup comparisons (see Table 2). For
example, the manual from the SB5 (Roid, 2003) reported
that children previously diagnosed with mental retardation
scored 2–3 SD below the normative mean on the SB5,
whereas children previously classified as gifted scored 1–2
SD above the normative mean.
Predictive ability. Predictive evidence, perhaps one of the
strongest forms of validity evidence, is reported least often.
Predictive validity evidence encompasses the effectiveness
of a test in predicting the performance of individuals in
related activities. For example, a measure might report the
extent to which its scores predict children’s educational
placement or later academic achievement. Obviously,
collecting such evidence requires laborious longitudinal
investigation with potentially unclear results. As such,
predictive evidence is provided by only 2 of the 16 measures, the K–ABC (Kaufman & Kaufman, 1983) and the
SPM–P (Raven et al., 1998). The K–ABC presents results
from six studies that used the K–ABC to predict scores on
achievement tests administered 6–12 months later. Similarly
the SPM–P reports correlations “ranging up to about .70”
between scores from the SPM–P and scores from achievement tests administered “some time” later (Raven et al.,
1998, p. 29). Additional measures, such as the SB5 (Roid,
2003) and UNIT (Bracken & McCallum, 1998), presented
correlation data with concurrently administered achievement
tests as evidence of predictive validity. However, the
concurrent administration of the tests disqualified it as
predictive evidence under our criteria.
Additional Resources
Given that for many measures our review represents but
one of many, we have provided the final column in Table 2
as a gateway to the research literature on these tests.
Although our list is not exhaustive, we have included all
reviews that we were able to access via PsycINFO as well
as those that were available through the Buros’s Mental
Measurements Yearbook series and the Tests in Print
series. In addition, we have included reviews from
Athanasiou (2000), Bracken and Naglieri (2003), and
McCallum (2003), as well as those cited within McGrew
and Flanagan (1998). Note that a limited number of
published reviews are available for the recently developed
or revised measures. Readers are referred to the Buros
Institute Mental Measurements Yearbook Test Reviews
On-Line at http://buros.unl.edu/buros/jsp/search.jsp and the
Educational Testing Service’s Test Link database available
at http://www.ets.org/testcoll/index.html for current
information.
Recommendations
Selection of an Appropriate Measure
Despite the responsibility incumbent on test developers,
“the ultimate responsibility for appropriate test use and
interpretation lies predominately with the test user”
(American Educational Research Association et al., 1999,
p. 111). We recognize that the administration of IQ
measures is primarily the responsibility of psychologists
rather than speech-language pathologists. However, we
provide suggestions for test selection for two primary
reasons. First, both clinicians and investigators sometimes
select and administer IQ measures within the scope of their
assessment protocols. Second, it is our hope that speechlanguage professionals will use the information presented
within this article to engage in meaningful dialogue with
collaborating psychologists regarding how IQ scores
should (and should not) be interpreted and which measures
are most appropriate for children with communication
difficulties. To this end, we offer three considerations to
guide the selection of a nonverbal IQ measure. First, the
measure should be psychometrically sound, as previously
discussed. Although none of the measures presented in
Table 2 are psychometrically ideal, four emerge as
particularly strong: the TONI–3 (Brown et al., 1997),
UNIT (Bracken & McCallum, 1998), WASI (Psychological Corporation, 1999), and WISC–IV (Wechsler, 2003).
Although all four measures received negative marks in
terms of test–retest reliability, all fell close to the .90
criterion (although note that the WASI and TONI–3
represented information on collapsed age groups only).
Similarly, the TONI–3 and UNIT received negative marks
under internal reliability, but their values fell just below the
.90 criterion, at .89 for certain age groups. In addition, the
UNIT received a negative mark under sample size, but that
was due specifically to the 16–17-year-old age grouping.
The second consideration for test selection is potential
special needs of the population or individual being evaluated. Given our particular interest in children with language difficulties, it is important to consider the linguistic
demands of individual IQ tests. Standard 7.7 (American
Educational Research Association et al., 1999) specifically
indicates, “in testing applications where the level of
DeThorne & Schaefer: Nonverbal IQ
285
linguistic or reading ability is not part of the construct of
interest, the linguistic or reading demands of the test
should be kept to the minimum necessary for the valid
assessment of the intended construct” (p. 82). Clearly, the
nonverbal instruments listed in Table 1 all attempt to
address this standard but with differing levels of success. If
concerns exist regarding language abilities—particularly
receptive language skills—test users should select a test
that involves nonverbal instructions only, such as the
UNIT (Bracken & McCallum, 1998) or the TONI–3
(Brown et al., 1997). Although the Leiter–R (Roid &
Miller, 1997) also relies on nonverbal instruction, its
psychometric properties do not appear as favorable as the
UNIT and the TONI–3. In terms of special needs, note also
that the color scheme of the UNIT (green and black) was
specifically designed to avoid most problems associated
with color blindness. For children with limited motor
control, examiners might want to give special consideration to the TONI–3, because it is untimed and does not
depend heavily on manipulatives.
The third factor to consider in selecting a nonverbal IQ
measure is the purpose of assessment. High-stakes assessments that will result in diagnoses, treatment eligibility,
and/or educational placement should use multidimensional
batteries, such as the UNIT (Bracken & McCallum, 1998)
or the WISC–IV (Wechsler, 2003). In contrast, if one is
conducting a low-stakes assessment, such as a screening
for research purposes, a one-dimensional battery may
suffice, such as the TONI–3 (Brown et al., 1997) or the
WASI (Psychological Corporation, 1999). McCallum et al.
(2001) describe one-dimensional measures as assessing “a
narrow aspect of intelligence through the use of progressive matrices,” whereas multidimensional measures “assess
multiple facets of examinees’ intelligence,” such as
memory, reasoning, and attention (pp. 9–10). In sum, given
all three aforementioned considerations, if one is selecting
a nonverbal IQ measure for a child with language impairment, we recommend the UNIT for cases of high-stakes
assessment and the TONI–3 for cases of low-stakes
assessment.
Interpretation of Nonverbal IQ Measures
Once an IQ measure has been selected and administered, one’s focus turns naturally to interpretation. IQ
scores as a whole are interpreted as estimates of an
individual’s current level of general cognitive functioning.
IQ scores tend to correlate positively with academic
achievement (r = .50–.75; Anastasi, 1988; Neisser et al.,
1996) and have proved to be one of the best predictors of
academic and job-related outcomes (Neisser et al., 1996;
Sattler, 2001). However, given our focus on nonverbal IQ
measures, it seems important to note that nonverbal IQ has
proven to be less effective in predicting academic outcomes than verbal IQ (Sattler, 2001). This difference in
predictive power is potentially due to two explanations,
neither of which precludes the other: (a) Verbal subtests
have a higher percentage of variance attributable to general
intelligence, or g, than do nonverbal subtests, and (b)
language plays an important role in the standard academic
286
learning environment. In regard to the latter point, it is
difficult for children with poor language skills in general to
succeed in school regardless of their level of general
cognitive functioning (Records, Tomblin, & Freese, 1992;
Tomblin, Zhang, Buckwalter, & Catts, 2000).
Regardless of predictive power, the most accurate
interpretation of an individual’s performance on any given
measure comes from an understanding of the specific
ability or abilities reflected by that measure. As previously
mentioned, factor analytic techniques have given shape to a
number of broad cognitive abilities (see Carroll, 1993;
Horn & Cattell, 1966). Each broad cognitive ability can be
further dissected into numerous narrow cognitive abilities,
a discussion of which is beyond the scope of this article
(see Carroll, 1993; McCallum, 2003; McGrew & Flanagan,
1998). McGrew and Flanagan have proposed the Cattell–
Horn–Carroll theory, which includes 10 broad cognitive
abilities: crystallized intelligence, fluid intelligence,
quantitative knowledge, reading and writing ability, shortterm memory, visual processing, auditory processing, longterm storage and retrieval, processing speed, and decision/
reaction time.
McGrew and Flanagan (1998) used the broad intelligences from the Cattell–Horn–Carroll theory as a framework for reviewing a number of IQ measures. They
concluded that nonverbal IQ measures as a whole draw
heavily on visual processing abilities and fluid intelligence
(see also analyses by Johnston, 1982, and Kamhi, Minor,
& Mauer, 1990). Fluid intelligence, typified by inductive
and deductive reasoning, refers to “mental operations that
an individual may use when faced with a relatively novel
task that cannot be performed automatically” (McGrew &
Flanagan, 1998, p. 14). Select examples of such tasks
include the Picture Similarities subtest from the DAS (C.
D. Elliott, 1990a); the Photo Series subtest from the K–
ABC (Kaufman & Kaufman, 1983); the Analogic Reasoning subtest from the UNIT (Bracken & McCallum, 1998);
and the matrices tasks that essentially define the NNAT
(Naglieri, 2003); CMMS–3 (Burgemeister et al., 1972);
CTONI (Hammill et al., 1997); SPM–P (Raven et al.,
1998); and TONI–3 (Brown et al., 1997).
Visual processing is defined by McGrew and Flanagan
(1998) as “the ability to generate, perceive, analyze,
synthesize, manipulate, transform, and think with visual
patterns and stimuli” (p. 23). Example measures of visual
processing taken from McGrew and Flanagan include all
the subsets from the K–ABC (Kaufman & Kaufman, 1983)
as well as the Block Building, Pattern Construction, and
Copying subtests from the DAS (C. D. Elliott, 1990a). One
would expect similar subtests across nonverbal IQ measures to be associated with visual processing ability as
well, such as the Design Analogies subtest from the Leiter–
R (Roid & Miller, 1997) and the Block Design subtest
from the WASI (Psychological Corporation, 1999). Note
that any single subtest can and likely does reflect more
than one broad cognitive ability. For example, in addition
to tapping visual processing skills, it seems likely that the
Cube Design subtest from the UNIT (Bracken &
McCallum, 1998) would be influenced by decision/
reaction time. Similarly, the Symbolic Memory and Spatial
American Journal of Speech-Language Pathology • Vol. 13 • 275–290 • November 2004
Memory subtests from the UNIT are likely to be influenced by both visual processing and short-term memory.
Table 3 contains proposed links between broad cognitive abilities from the Cattell–Horn–Carroll theory and
subtests from the four measures that emerged from our
review as psychometrically strong: the TONI–3 (Brown et
al., 1997); UNIT (Bracken & McCallum, 1998); WASI
(Psychological Corporation, 1999); and WISC–IV
(Wechsler, 2003). Such proposed links are based largely on
the work of McCallum (2003) and McGrew and Flanagan
(1998), with some additional speculation on our part. Note
that although individual subtests may tap into different
nonverbal abilities, subtests’ often inferior psychometric
properties make interpretation in isolation or in comparison
with other subtests largely speculative (see McDermott &
Glutting, 1997). Consequently, reliability coefficients for
the individual subtests are provided in Table 3 for reference purposes.
The published test reviews cited in Table 2 revealed
little additional information on the broad cognitive
requirements of individual IQ measures, although we
summarize here what we found for our two recommended
measures: the TONI–3 and the UNIT. Both have been
criticized for providing a rather limited assessment of more
specific cognitive abilities. For example, the TONI–3 has
been criticized for its one-dimensional nature (Athanasiou,
2000; see also Kamhi et al., 1990). Along similar lines,
Atlas (2001, p. 1260) noted that the TONI–3’s correlation
with the widely accepted and multidimensional WISC–III
was “at best moderate” (r = .53–.63) and based on a small
sample (i.e., 34 students). In contrast, however, DeMauro
(2001) attributed this attenuated correlation to the higher
verbal demands of the Weschler scales and, as such,
presented it as rather positive evidence of the TONI–3’s
validity. In regard to the UNIT (Bracken & McCallum,
1998), Young and Assing (2000) credited its manual with a
thorough description of the test’s theoretical foundation,
but Fives and Flanagan (2002) criticized the measure for
not providing a clear and comprehensive association with
more specific cognitive abilities. As an example, Fives and
Flanagan (2002) suggested that an examinee’s low scores
on the Symbolic Memory and Spatial Memory subtests
could be attributed to “limitations in memory, reasoning, or
both of these constructs” (p. 430). In closing, regardless of
whether one is interpreting individual subtest or composite
scores, it is important to bear in mind that valid assessment
of cognition, as with any developmental domain, must
combine information from multiple sources, including
observation and interview.
Summary
In sum, the present article presented information
regarding 16 commonly used child nonverbal IQ measures,
highlighted important distinctions among them, and
provided recommendations for test selection and interpretation. When selecting a nonverbal IQ measure for children
with language difficulties, we currently recommend the
UNIT (Bracken & McCallum, 1998) for cases of highstakes assessment and the TONI–3 (Brown et al., 1997) for
cases of low-stakes assessment. Regardless of which
specific test is administered, professionals are encouraged
to be cognizant of each test’s strengths and limitations,
including its psychometric properties and its potential
TABLE 3. Broad cognitive abilities and reliability coefficients associated with individual subtests from the recommended IQ
measures.
IQ test
Test of Nonverbal Intelligence—Third
Edition (TONI–3; Brown, Sherbenou,
& Johnsen, 1997)
Universal Nonverbal Intelligence Test
(UNIT; Bracken & McCallum, 1998)
Cognitive abilitya
rb
Abstract Figural Problem-Solving
Fluid intelligence
.89–94c
Symbolic Memory
Visual processing
Short-term memory
Visual processing
Short-term memory
Visual processing
Fluid intelligence
Decision/reaction time
Fluid intelligence
Visual processing
Fluid intelligence
Decision/reaction time
Fluid intelligence
Decision/reaction time
Visual processing
Fluid intelligence
Decision/reaction time
Fluid intelligence
Visual processing
.80–87
Nonverbal subtest
Spatial Memory
Cube Design
Wechsler Abbreviated Scale of Intelligence
(WASI; Psychological Corporation, 1999)
Analogic Reasoning
Block Design
Matrix Reasoning
Wechsler Intelligence Scale for Children—
Fourth Edition (WISC–IV;
Wechsler, 2003)
Block Design
Picture Concepts
Matrix Reasoning
.74–.84
.81–.95
.70–.83
.84–.93
.86–.96
.83–.88
.76–.85
.86–.92
a
Broad cognitive abilities from the Cattell–Horn–Carroll theory, as proposed by McGrew and Flanagan (1998). bRange of subtest reliability
coefficients across ages for that particular subtest. cSubtest reliability for the TONI–3 is the same as for the entire test.
DeThorne & Schaefer: Nonverbal IQ
287
relation to different cognitive abilities. We hope that this
guide of nonverbal intelligence measures will help clinicians and investigators to better interpret available IQ
scores and, when applicable, select the most appropriate
instrument for their specific needs.
Acknowledgments
We wish to thank John McCarthy, Joanna Burton, Patricia
Ruby, and Colleen Curran for their assistance with this project.
References
American Educational Research Association, American
Psychological Association, and National Council on
Measurement in Education. (1999). Standards for educational and psychological testing. Washington, DC: American
Psychological Association.
Anastasi, A. (1985). Review of the Kaufman Assessment Battery
for Children. In J. V. Mitchell (Ed.), Mental measurements
yearbook (9th ed., pp. 769–771). Lincoln, NE: Buros Institute
of Mental Measurements.
Anastasi, A. (1988). Psychological testing (6th ed.). New York:
Macmillan.
Athanasiou, M. S. (2000). Current nonverbal assessment
instruments: A comparison of psychometric integrity and test
fairness. Journal of Psychoeducational Assessment, 18,
211–229.
Athanasiou, M. (2003). Review of the Pictorial Test of
Intelligence, Second edition. In B. S. Plake, J. C. Impara, &
R. A. Spies (Eds.), Mental measurements yearbook (15th ed.,
pp. 676–678). Lincoln, NE: Buros Institute of Mental
Measurements.
Atlas, J. A. (2001). Review of the Test of Nonverbal Intelligence,
Third edition. In B. S. Plake, & J. C. Impara (Eds.), Mental
measurements yearbook (14th ed., pp. 1259–1260). Lincoln,
NE: Buros Institute of Mental Measurements.
Aylward, G. P. (1992). Review of the Differential Ability Scales.
In J. J. Kramer & J. C. Conoley (Eds.), Mental measurements
yearbook (11th ed., pp. 281–282). Lincoln, NE: Buros
Institute of Mental Measurements.
Aylward, G. P. (1998). Review of the Comprehensive Test of
Nonverbal Intelligence. In J. C. Impara & B. S. Plake (Eds.),
The mental measurements yearbook (13th ed., pp. 310–312).
Lincoln, NE: Buros Institute of Mental Measurements.
Bandalos, D. L. (2001). Review of the Universal Nonverbal
Intelligence Test. In B. S. Plake & J. C. Impara (Eds.), Mental
measurements yearbook (14th ed., pp. 1296–1298). Lincoln,
NE: Buros Institute of Mental Measurements.
Bortner, M. (1965). Review of the Raven Standard Progressive
Matrices. In O. K. Buros (Ed.), Mental measurements
yearbook (6th ed., pp. 489–491). Highland Park, NJ: Gryphon
Press.
Bracken, B. A. (1985). A critical review of the Kaufman Battery
for Children (K–ABC). School Psychology Review, 14, 21–36.
Bracken, B. A. (1987). Limitations of preschool instruments and
standards for minimal levels of technical adequacy. Journal of
Psychoeducational Assessment, 4, 313–326.
Bracken, B. A. (1988). Ten psychometric reasons why similar
tests produce dissimilar results. Journal of School Psychology,
26, 155–166.
Bracken, B. A. (Ed.). (2000). The psychoeducational assessment
of preschool children (3rd ed.). Boston: Allyn & Bacon.
Bracken, B. A., & McCallum, R. S. (1998). Universal Nonverbal Intelligence Test. Itasca, IL: Riverside Publishing.
288
Bracken, B. A., & Naglieri, J. A. (2003). Assessing diverse
populations with nonverbal tests of general intelligence. In C.
R. Reynolds & R. W. Kamphaus (Eds.), Handbook of
psychological and educational assessment of children:
Intelligence, aptitude, and achievement (2nd ed., pp. 243–
274). New York: Guilford Press.
Bradley-Johnson, S. (1997). Review of the Comprehensive Test
of Nonverbal Intelligence. Psychology in the Schools, 34,
289–292.
Brown, L., Sherbenou, R. J., & Johnsen, S. K. (1997). Test of
Nonverbal Intelligence—Third Edition. Austin, TX: Pro-Ed.
Burgemeister, B. B., Hollander, L. H., & Lorge, I. (1972).
Columbia Mental Maturity Scale—Third Edition. San
Antonio, TX: Psychological Corporation.
Carroll, J. B. (1993). Human cognitive abilities: A survey of
factor-analytic studies. Cambridge, England: Cambridge
University Press.
Casby, M. W. (1992). The cognitive hypothesis and its influence
on speech–language services in schools. Language, Speech,
and Hearing Services in Schools, 23, 198–202.
Castellanos, M., Kline, R. B., & Snyder, J. (1996). Lessons
from the Kaufman Assessment Battery for Children (K–
ABC): Toward a new cognitive assessment model. Psychological Assessment, 8, 7–17.
Coffman, W. E. (1985). Review of the Kaufman Assessment
Battery for Children. In J. V. Mitchell (Ed.), Mental measurements yearbook (9th ed., pp. 771–773). Lincoln, NE: Buros
Institute of Mental Measurements.
Conoley, J. C. (1990). Review of the K–ABC: Reflecting the
unobservable. Journal of Psychoeducational Assessment, 8,
369–375.
DeMauro, G. E. (2001). Review of the Test of Nonverbal
Intelligence, Third edition. In B. S. Plake & J. C. Impara
(Eds.), Mental measurements yearbook (14th ed., pp. 1260–
1262). Lincoln, NE: Buros Institute of Mental Measurements.
Drossman, E. R., & Maller, S. J. (2000). Review of the
Comprehensive Test of Nonverbal Intelligence. Journal of
Psychoeducational Assessment, 18, 293–301.
Egeland, B. R. (1978). Review of the Columbia Mental Maturity
Scale. In O. K. Buros (Ed.), Mental measurements yearbook
(8th ed., pp. 298–299). Highland Park, NJ: Gryphon Press.
Elliott, C. D. (1990a). Differential Ability Scales. San Antonio,
TX: Psychological Corporation.
Elliott, C. D. (1990b). The nature and structure of children’s
abilities: Evidence from the Differential Ability Scales.
Journal of Psychoeducational Assessment, 8, 376–390.
Elliott, S. N. (1990). The nature and structure of the DAS:
Questioning the test’s organizing model and use. Journal of
Psychoeducational Assessment, 8, 406–411.
Farrell, M., & Phelps, L. (2000). A comparison of the Leiter–R
and the Universal Nonverbal Intelligence Test (UNIT) with
children classified as language impaired. Journal of Psychoeducational Assessment, 18, 268–274.
Fives, C. J., & Flanagan, R. (2002). A review of the Universal
Nonverbal Intelligence Test (UNIT): An advance for evaluating youngsters with diverse needs. School Psychology
International, 23, 425–449.
Flanagan, D., & Alfonso, V. (1995). A critical review of the
technical characteristics of new and recently revised intelligence tests for preschool children. Journal of Pyschoeducational Assessment, 13, 66–90.
Flanagan, D. P., & Caltabiano, L. F. (2003). Review of the
Pictorial Test of Intelligence, Second edition. In B. S. Plake, J.
C. Impara, & R. A. Spies (Eds.), Mental measurements
yearbook (15th ed., pp. 678–681). Lincoln, NE: Buros
Institute of Mental Measurements.
American Journal of Speech-Language Pathology • Vol. 13 • 275–290 • November 2004
Flynn, J. R. (1999). Searching for justice: The discovery of IQ
gains over time. American Psychologist, 54, 5–20.
French, J. L. (2001). Pictorial Test of Intelligence—Second
Edition. Austin, TX: Pro-Ed.
Glutting, J., Adams, W., & Sheslow, D. (2000). Wide Range
Intelligence Test. Wilmington, DE: Wide Range.
Hamilton, W., & Burns, T. G. (2003). WPPSI–III: Wechsler
Preschool and Primary Scale for Intelligence (3rd ed.).
Applied Neuropsychology, 10, 188–190.
Hammill, D. D., Pearson, N. A., & Wiederholt, J. E. (1997).
Comprehensive Test of Nonverbal Intelligence. Austin, TX:
Pro-Ed.
Hessler, G. (1985). Review of the Kaufman Assessment Battery
for Children: Implications for assessment of the gifted.
Journal for the Education of the Gifted, 8, 133–147.
Hopkins, K. D., & Hodge, S. E. (1984). Review of the Kaufman
Assessment Battery (K–ABC) for Children. Journal of
Counseling and Development, 63, 105–107.
Horn, J. L., & Cattell, R. B. (1966). Refinement and test of the
theory of fluid and crystallized general intelligences. Journal
of Educational Psychology, 57, 253–270.
Johnston, J. R. (1982). Interpreting the Leiter IQ: Performance
profiles of young normal and language-disordered children.
Journal of Speech and Hearing Research, 25, 291–296.
Kamhi, A. G., Minor, J. S., & Mauer, D. (1990). Content
analysis and intratest performance profiles on the Columbia
and the TONI. Journal of Speech and Hearing Research, 33,
375–379.
Kamphaus, R. W. (1990). K–ABC theory in historical and
current contexts. Journal of Pyschoeducational Assessment, 8,
356–368.
Kamphaus, R. W., & Reynolds, C. R. (1984). Development and
structure of the Kaufman Assessment Battery for Children (K–
ABC). Journal of Special Education, 18, 213–228.
Kaufman, A. S. (1978). Review of the Columbia Mental
Maturity Scale. In O. K. Buros (Ed.), Mental measurements
yearbook (8th ed., pp. 299–301). Highland Park, NJ: Gryphon
Press.
Kaufman, A. S., & Kaufman, N. L. (1983). K–ABC: Kaufman
Assessment Battery for Children. Circle Pines, MN: American
Guidance Service.
Keith, T. Z. (1985). Questioning the K–ABC: What does it
measure? School Psychology Review, 14, 9–20.
Keith, T. Z. (2001). Review of the Wechsler Abbreviated Scale
of Intelligence. In B. S. Plake & J. C. Impara (Eds.), Mental
measurements yearbook (14th ed., pp. 1329–1331). Lincoln,
NE: Buros Institute of Mental Measurements.
Linkskog, C. O., & Smith, J. V. (2001). Review of the Wechsler
Abbreviated Scale of Intelligence. In B. S. Plake & J. C.
Impara (Eds.), Mental measurements yearbook (14th ed., pp.
1331–1332). Lincoln, NE: Buros Institute of Mental
Measurements.
Marco, G. L. (2001). Review of the Leiter International
Performance Scale—Revised. In B. S. Plake & J. C. Impara
(Eds.), Mental measurements yearbook (14th ed., pp. 683–
687). Lincoln, NE: Buros Institute of Mental Measurements.
McCallum, R. S. (2003). Handbook of nonverbal assessment.
New York: Kluwer Academic/Plenum.
McCallum, R. S., Bracken, B., & Wasserman, J. (2001).
Essentials of nonverbal assessment. New York: Wiley.
McCauley, R. J. (2001). Assessment of language disorders in
children. Mahwah, NJ: Erlbaum.
McCauley, R. J., & Swisher, L. (1984a). Psychometric review
of language and articulation tests for preschool children.
Journal of Speech and Hearing Disorders, 49, 34–42.
McCauley, R. J., & Swisher, L. (1984b) Use and misuse of
norm-referenced tests in clinical assessment: A hypothetical
case. Journal of Speech and Hearing Disorders, 49, 338–348.
McDermott, P. A., & Glutting, J. J. (1997). Informing stylistic
learning behavior, disposition, and achievement through
ability subtests—Or, more illusions of meaning? School
Psychology Review, 26, 163–175.
McFadden, T. U. (1996). Creating language impairments in
typically achieving children: The pitfalls of “normal”
normative sampling. Language, Speech, and Hearing Services
in Schools, 27, 3–9.
McGrew, K. S., & Flanagan, D. P. (1998). The intelligence test
desk reference: Gf–Gc cross-battery assessment. Boston:
Allyn & Bacon.
Mehrens, W. A. (1984). A critical analysis of the psychometric
properties of the K–ABC. Journal of Special Education, 18,
297–310.
Messick, S. (1989). Validity. In R. L. Linn (Ed.), Educational
measurement (3rd ed., pp. 13–109). New York: American
Council on Education and Macmillan.
Naglieri, J. A. (2003). Naglieri Nonverbal Ability Test. San
Antonio, TX: Psychological Corporation.
Naglieri, J. A., & Das, J. P. (1997). Cognitive Assessment
System. Itasca, IL: Riverside.
Neisser, U., Boodoo, G., Bouchard, T. J., Boykin, A. W.,
Brody, N., Ceci, S. J., et al. (1996). Intelligence: Knowns and
unknowns. American Psychologist, 51, 77–101.
Page, E. B. (1985). Review of the Kaufman Assessment Battery
for Children. In J. V. Mitchell (Ed.), Mental measurements
yearbook (9th ed., pp. 773–777). Lincoln, NE: Buros Institute
of Mental Measurements.
Plante, E. (1998). Criteria for SLI: The Stark and Tallal legacy
and beyond. Journal of Speech, Language, and Hearing
Research, 41, 951–957.
Platt, L. O., Kamphaus, R. W., Keltgen, J., & Gilliland, F.
(1991). An overview and review of the Differential Ability
Scales: Initial and current research findings. Journal of School
Psychology, 29, 271–277.
Psychological Corporation. (1999). Wechsler Abbreviated Scale
of Intelligence. San Antonio, TX: Author.
Raven, J., Raven, J. C., & Court, J. H. (1998). Standard
Progressive Matrices–Parallel form. Oxford, England: Oxford
Psychologists Press.
Records, N. L., Tomblin, J. B., & Freese, P. R. (1992). The
quality of life of young adults with histories of specific
language impairment. American Journal of Speech-Language
Pathology, 1, 44–53.
Reinehr, R. C. (1992). Review of the Differential Ability Scales.
In J. J. Kramer & J. C. Conoley (Eds.), Mental measurements
yearbook (11th ed., pp. 282–283). Lincoln, NE: Buros
Institute of Mental Measurements.
Reynolds, C. R., & Kamphaus, R. W. (2003). Reynolds
Intellectual Assessment Scales. Lutz, FL: Psychological
Assessment Resources.
Roid, G. H. (2003). Stanford–Binet Intelligence Scale—Fifth
Edition. Itasca, IL: Riverside.
Roid, G. H., & Miller, L. J. (1997). Leiter International
Performance Scale—Revised. Wood Dale, IL: Stoelting.
Salvia, J., & Ysseldyke, J. E. (1998). Assessment (7th ed.).
Boston: Houghton Mifflin.
Sattler, J. M. (2001). Assessment of children: Cognitive
applications (4th ed.). San Diego, CA: Author.
Stark, R. E., & Tallal, P. (1981). Selection of children with
specific language deficits. Journal of Speech and Hearing
Disorders, 46, 114–122.
Sternberg, R. J. (1983). Should K come before A, B, and C? A
review of the Kaufman Assessment Battery for Children
DeThorne & Schaefer: Nonverbal IQ
289
(K–ABC). Contemporary Education Review, 2, 199–207.
Sternberg, R. J. (1984). The Kaufman Assessment Battery for
Children: An information-processing analysis and critique.
Journal of Special Education, 18, 269–279.
Stinnett, T. A. (2001). Review of the Leiter International
Performance Scale–Revised. In B. S. Plake & J. C. Impara
(Eds.), Mental measurements yearbook (14th ed., pp. 687–
692). Lincoln, NE: Buros Institute of Mental Measurements.
Stinnett, T. A. (2003). Review of the Wide Range Intelligence
Test. In B. S. Plake, J. C. Impara, & R. A. Spies (Eds.),
Mental measurements yearbook (15th ed., pp. 1011–1015).
Lincoln, NE: Buros Institute of Mental Measurements.
Tomblin, J. B., Zhang, X., Buckwalter, P., & Catts, H. (2000).
The association of reading disability, behavioral disorders, and
language impairment among second-grade children. Journal
of Child Psychology and Psychiatry and Allied Disciplines,
41, 473–482.
U.S. Bureau of the Census. (2001). Census 2000 Summary File
1—United States [Machine-readable computer file]. Washington, DC: Author.
Van Lingen, G. (1998). Review of the Comprehensive Test of
Nonverbal Intelligence. In J. C. Impara & B. S. Plake (Eds.),
The mental measurements yearbook (13th ed., pp. 312–314).
Lincoln, NE: Buros Institute of Mental Measurements.
Wechsler, D. (2002). Wechsler Preschool and Primary Intelligence Scale—Third Edition. San Antonio, TX: Psychological
Corporation.
290
Wechsler, D. (2003). Wechsler Intelligence Scale for Children—
Fourth Edition. San Antonio, TX: Psychological Corporation.
Whitmire, K. (2000). Cognitive referencing and discrepancy
formulae: Comments from ASHA resources. ASHA Special
Interest Division 1, Language Learning and Education, 7,
13–16.
Widaman, K. F. (2003). Review of the Wide Range Intelligence
Test. In B. S. Plake, J. C. Impara, & R. A. Spies (Eds.),
Mental measurements yearbook (15th ed., pp. 1015–1017).
Lincoln, NE: Buros Institute of Mental Measurements.
Woodcock, R. W., McGrew, K. S., & Mather, N. (2001).
Woodcock–Johnson Tests of Cognitive Abilities—III. Itasca,
IL: Riverside.
Young, E. L., & Assing, R. (2000). Review of the Universal
Nonverbal Intelligence Test. Journal of Psychoeducational
Assessment, 18, 280–288.
Received April 5, 2004
Revision received June 30, 2004
Accepted October 7, 2004
DOI: 10.1044/1058-0360(2004/029)
Contact author: Laura S. DeThorne, PhD, Pennsylvania State
University, 110 Moore Building, University Park, PA 16802.
E-mail: lsd12@psu.edu
American Journal of Speech-Language Pathology • Vol. 13 • 275–290 • November 2004
Download